MSR MAP Demo
Privacy policy
Contact us

*Heritage Health Prize

Heritage Provider Network (HPN) is offering The Heritage Health Prize, $3.3 million to whoever develops a breakthrough algorithm that uses available patient data to predict and hopefully prevent future hospitalizations. HPN is sponsoring the Heritage Health Prize Competition. HPN firmly believes that competition is the best way to achieve the monumental breakthroughs necessary to fix America’s health care system. The hoped-for algorithm will be used to predict how many days a patient will spend in the hospital in the next year. Health care providers will use this to develop new care plans and strategies to reach patients before emergencies can occur. It is hoped that this will thereby reduce a significant number of unnecessary hospitalizations. This will ideally result in increasing the health of patients while decreasing the cost of care.

Medical Scientific Resources was initially quite pleased to learn about the Heritage Health Prize competition. The intent of the contest is admirable because it addresses a serious issue – the cost of health care.

However, after reviewing the contest rules and the description of the data set content, we realize that the Heritage Health Prize contest predictive model won’t work to achieve the stated goal, to reduce health care costs by reducing unnecessary hospital admissions and days. The duration of the prediction is simply too long. In addition, there are important, critical variables not included in their model which affect future hospitalization and hospital stays.

Critical information not available in the provided (and anticipated) data sets from the Heritage Health Plan include variables which significantly impact future hospitalization of a patient. Some of the missing data elements relate to the following behaviors of physicians, hospitals and patients:

(1) Many patients don’t receive necessary outpatient care, medicines or follow- up care because they can’t afford it;
(2) Physician admission decisions are often based on non-recorded significant patient medical history and physical examination findings – simply put, the reason(s) why a patient was admitted aren’t always recorded in the medical record or data set;
(3) Many patients’ health condition deteriorates because they don’t follow physician instructions consistently, and/or they don’t have the social support network necessary to help them when they are sick;
(4) Many patients have destructive lifestyles which result in unanticipated injuries or illnesses requiring hospitalization; and
(5) Many physicians and hospitals fail to effectively communicate to their patients what they need to do after discharge, resulting in readmissions.

We have previously studied this issue and found that, in order to make a predictive model actually “work”, we needed to set the predictive time length to zero and include additional variables in the data set. Because these variables are truly dynamic, and can constantly change in each patient’s life and circumstances, a real-time approach is essential.

Our Monitor to Assess Progress predictive model is used as a guide by the physician to help determine if an admission is appropriate that very day, or determine if the hospitalized patient needs to continue the stay or can be safely discharged that very day. When using our model, as confirmed in our beta testing, the goal of the Heritage Health Prize is achieved – dramatic reductions in admission rates, hospital days and related costs are realized.

We are pleased with the results our software has provided, and we look forward to working with our clients to effectively control admission rates and related hospital costs.

Respectfully, MEDICAL SCIENTIFIC RESOURCES OF NEVADA, INC.

Kenneth Osgood, MD, MPH, President